Intel's 10nm only coming to servers in 2020 with Ice Lake

[leadeater]

@AngryBeaver

Think you might have some miss-quotes there, the ones quoted as @cj09beira I'm very sure are quotes of quotes so show up as the wrong person.

[Taf the Ghost]

30 minutes ago, TechyBen said:

Interesting there is no interim product? 12++ or 10nm with some 12nm mixed in?

14nm+++ is a 12nm product in other stacks.

[yian88]

2 hours ago, leadeater said:

So I need to buy a dGPU because I need a workstation for my job that has 6-8 cores and 32GB ram solely for running VMs and do not require any kind of GPU workloads at all other than driving a display? Sounds incredibly wasteful to me. Intel finally releases a platform that can meet these requirements that isn't HEDT and now you want to remove it as an option again? We'd love to reduce the cost of the computers we buy.

 

Or that software developer that totally needs a dGPU as well, or a business analyst running complex reporting on very large data sets.

 

There are a lot of office workers out there that don't need dGPUs or HEDT platforms that run more than Word and check emails.

 

Cheap dGPUs are not a good solution, they fail a lot and it's just another thing to go wrong in a computer that doesn't require one.

 

Business market is just so much bigger than gaming so those requirements come first, then you get K SKUs from those products.

 

Nobody reads what im writing just biased people, i keep saying they can continue to manufacture laptop and lower end desktop chips with iGPU to counter AMD and market needs. AMD does not have high end APU, none of the Ryzen chips have iGPU so intel removing iGPU from i5/i7 chips doesnt make AMD have a monopoly in that area.

Just remove them from high end desktop CPU, instead of making only 1 size fits all product, they should manufacture 2 product ranges but they wont so they can sell you useless silicon.

You are missing the point, you are paying a certain ammount of money for the iGPU which could either be removed and have cheaper CPU's or it could be replaced for more cores for the same price which we really need.

 

Maybe words is not enough, maybe a picture can help you realize how much die space is being wasted which you are paying for, which is the main reason AMD can deliver 8 core for less $ than intel 6 core.

What do you see on the green area on the right? i define it as 50-100$ wasted die space/ wafer space per processor or 2 extra cores and more L3 cache for the same price.

Hypocrites, i see people doing so much wafer maths on articles, videos and forums, on yields and how important die area mm2 really is, but when it comes to intels 1/3 of CPU being wasted die, hence 1/3 of all wafer space for high end chips is being payed for unwilingly by just being forced to buy intel cpu with iGPU.

If an intel wafer with 100 i7 chips costs 20 000$  1/3 or 6000$ is being wasted away, end price on shelves might be lower ratio like 15-20% of the final price, but wafer alone thats my silly math.

 

Thats why i hope their business model 1 size fits all goes bust in face of AMD's upcoming Zen 2 chips with rumoured 16 cores on desktop. Im no fanboy i just know how much intel has been holding back from upgrading core count, first they locked all CPU's from OC then forced iGPU on all chips instead of extra cores and people bought into that crap.

I prefer the old model where certain motherboards had iGPU, if you wanted a simple iGPU buy a motherboard with iGPU, if you dont want buy a cheaper one without instead of everyone buying intel forced to pay for intel's gpu division.

 

 

 

Im done with this topic, no logic found here. I dont give a shit about macbooks or laptop's, i only use high end desktop PC, they need to change their 1 size fits all model, i dont need to pay for iGPU just so that laptop/macbooks can have one too.

I just hope AMD Zen2 gets some good IPC/ clock/cores gains next year so i can make the switch, unless intel changes their model, if the rumoured i9 8c/16t K sku on mainstream platform comes with iGPU it will be clear intel has no intention of delivering a good bang for buck and stop screwing customers.

[AngryBeaver]

15 minutes ago, leadeater said:

@AngryBeaver

Think you might have some miss-quotes there, the ones quoted as @cj09beira I'm very sure are quotes of quotes so show up as the wrong person.

Yes it is, I fixed the others. Looks like I missed one lol.

[leadeater]

6 minutes ago, yian88 said:

Nobody reads what im writing just biased people, i keep saying they can continue to manufacture laptop and lower end desktop chips with iGPU to counter AMD and market needs.

 

6 minutes ago, yian88 said:

Just remove them from high end desktop CPU, instead of making only 1 size fits all product, they should manufacture 2 product ranges but they wont so they can sell you useless silicon.

You are missing the point, you are paying a certain ammount of money for the iGPU which could either be removed and have cheaper CPU's or it could be replaced for more cores for the same price which we really need.

 

I did and I explained rather clearly why an i7 should have an iGPU, for the many thousands, millions, of computers that need that mount of CPU power and don't need a dGPU or want a dGPU in the system at all.

 

Your point isn't unheard or even necessarily not agreed with however there are more use cases for i7's with iGPUs than there are without, sure for gaming it makes a ton of sense but Intel isn't going to cater an entire platform to such a small market when there is a much bigger one that wants the iGPU in i7's.

 

6 minutes ago, yian88 said:

Just remove them from high end desktop CPU, instead of making only 1 size fits all product, they should manufacture 2 product ranges but they wont so they can sell you useless silicon.

So the Intel HEDT product line?

[Drak3]

8 minutes ago, leadeater said:

I did and I explained rather clearly why an i7 should have an iGPU, for the many thousands, millions, of computers that need that mount of CPU power and don't need a dGPU or want a dGPU in the system at all.

An alernate solution would be having SKUs with and sans iGPU.

[79wjd]

5 minutes ago, Drak3 said:

An alernate solution would be having SKUs with and sans iGPU.

It's probably (id be a bit amazed if it wasn't) cheaper to make one chip (with lower tooling/manufacturing/maintenance/space costs) and just disable the iGPU. 

[leadeater]

Just now, Drak3 said:

An alernate solution would be having SKUs with and sans iGPU.

If Intel thought they would sell enough to justify creating that die I'm sure they would of, though there is a much better chance of that happening now since Zen than ever before. However I see that more stepping on the HEDT product line, maybe they'll fold that in a bit and move the PCIe reduced HEDT SKUs (7800/7820) in to the no iGPU die with more cores and for the higher end full PCIe high core count put that under the Xeon W product line (already is anyway, i9 is just no ECC).

 

I've always thought the reduced PCIe HEDT CPUs were totally stupid so that is an area where I would want Intel to merge with the desktop product line, at least that way you'd have higher core count CPUs with a few more PCIe lanes in the fast architecture development cycle.

[yian88]

1 minute ago, leadeater said:

 

 

I did and I explained rather clearly why an i7 should have an iGPU, for the many thousands, millions, of computers that need that mount of CPU power and don't need a dGPU or want a dGPU in the system at all.

 

Your point isn't unheard or even necessarily not agreed with however there are more use cases for i7's with iGPUs than there are without, sure for gaming it makes a ton of sense but Intel isn't going to cater an entire platform to such a small market when there is a much bigger one that wants the iGPU in i7's.

This isnt about gaming, its about 2 different products intel should be making instead of 1, people needing an iGPU should have to pay for it and those buying just CPU should get cheaper chips or more cores, they get away with it because no one cares or knows the real cost of iGPU on final price, if you judge it by wafer/die area space its quite shocking, enough to stop being ignorant about what they are actually selling you.

And no having 2 product lines wouldnt increase cost, intel already has multiple fabs making various chip types, its just more profitable for them to sell that 30% more wafer silicon to you which could be used for more chips/wafer aka higher yields= lower prices, or add cores, maintain die size mm2 and have same price, they are sandbagging performance by witholding new product line of high end chips without iGPU and more cores, am i not making sense.

 

Its like buying a car with leather seats paying 2k$ more for the feature because 70% of the market uses/demands leather seats hence the other 30% should pay for leather seats aswell so the manufacturers can rip the 30% off by not offering a cheaper alternative.

 

If AMD would add vega iGPU to Ryzen 7 2700x its price with jump by 100$ or more and make all their yields fucked up due to the large space gpu takes on each chip, since they use same core dies on EPYC, Threadripper, Ryzen with infinity fabric, and new Zen APU's line with Vega is done separately.

Looking at the die indeed that would be ridiculous to add to Ryzen 5/7 6-8 core chips as no one would buy Ryzen with 100$ higher price, it only makes sense in lower end SKU's with 4 cores for people who dont want dGPU.

Ryzen 5 2400G.

[leadeater]

4 minutes ago, yian88 said:

This isnt about gaming, its about 2 different products intel should be making instead of 1, people needing an iGPU should have to pay for it and those buying just CPU should get cheaper chips or more cores, they get away with it because no one cares or knows the real cost of iGPU on final price, if you judge it by wafer/die area space its quite shocking, enough to stop being ignorant about what they are actually selling you.

And no having 2 product lines wouldnt increase cost, intel already has multiple fabs making various chip types, its just more profitable for them to sell that 30% more wafer silicon to you which could be used for more chips/wafer aka higher yields= lower prices, or add cores, maintain die size mm2 and have same price, they are sandbagging performance by witholding new product line of high end chips without iGPU and more cores, am i not making sense.

Thing is everything you are asking for or explaining already exists, Intel HEDT. You don't have to go crazy and buy a 7980XE and a Rampage board, there are much cheaper options and CPU wise 7800X/7820X/7900X are not crazy prices. Higher than I'd like to see but Intel already does what you want, maybe just not at the price you want to pay?

[Drak3]

5 minutes ago, leadeater said:

If Intel thought they would sell enough to justify creating that die I'm sure they would of, though there is a much better chance of that happening now since Zen than ever before.

Given how many OEM machines ship with GPUs now, that can now support things like 4K netflix, the iGPU is starting to just take up resources unnecessarily on most lower mid tier to high end systems that don't benefit from iGPUs.

[yian88]

19 minutes ago, leadeater said:

Thing is everything you are asking for or explaining already exists, Intel HEDT. You don't have to go crazy and buy a 7980XE and a Rampage board, there are much cheaper options and CPU wise 7800X/7820X/7900X are not crazy prices. Higher than I'd like to see but Intel already does what you want, maybe just not at the price you want to pay?

It doesnt exist,  6 core HEDT i7 7800X socket 2066 is slightly more expensive than socket 1151v2 i7 8700k, because intel wont cannibalize their mainstream platform by selling i7 7800x cheaper than 8700k so you can figure out the iGPU is actually quite an expensive piece.

Add the fact that the cheapest 2066 board i can find is 2-3x the price of a cheap b350 1151v2 board.

 

Basically if i want a performant i5 8 core unlocked without iGPU and fancy HEDT features like quad channel RAM, and dual GBE ports (which is overkill for regular user) intel will never give that to the market, they are sandbagging performance as i said. They will keep this model for as long as possible, unless AMD can deliver some serious competition with 8-16 cores on AM4 socket then i hope intel might have no choice but to deliver high end sku's i5/i7/i9 with 8-10-12 cores by removing iGPU, we will see when new i9 for 1151v2(or v3?) comes out, thats when i will know not even over their dead company they will change this scam business model ( i think so).

[leadeater]

1 minute ago, Drak3 said:

Given how many OEM machines ship with GPUs now, that can now support things like 4K netflix, the iGPU is starting to just take up resources unnecessarily on most lower mid tier to high end systems that don't benefit from iGPUs.

Business sector most don't, most computers are also SFF with bad airflow and half height slots. Last time I checked we had around 8000 SFF desktops with no dGPU in them and we don't allow one to be picked when ordering. There's a full tower option that only comes with a dGPU but we have far less of those.

 

I have no use for a dGPU but my entire team has to put up with buying super high end workstations just to get more than 4 cores and a useless quadro (that's an internal ordering issue for that though). Now that the 8700 is a thing we can get exactly what we want, 6 cores 32GB ram SFF chassis (desk space) for half what we paid last time round. None of us want a dGPU.

 

So I don't disagree with more cores being great on desktop but I do disagree that most i5/i7 systems don't want an iGPU in them. If there was such a big demand for more cores and no iGPU that couldn't be catered for on HEDT then Intel would have done something about it, if there is money to be made Intel would go after it. More put up with the desktop product stack and it's limitations than step up to HEDT, very few business desktops come with that configuration.

 

I'd rather our room full of HPE Z series workstations turn in to HPE EliteDesk SFF desktops with no dGPU than have a low grade dGPU in the system that will fail in an office environment. I've seen enough fail in the older 7000/8000 Elite SFF series to know it's more a certainty than a maybe.

 

I can tell you one thing that is true across the world, building space therefore desk space/work space costs many times more than the iGPU in an Intel CPU so taking out the iGPU and putting in dGPUs would cost more if you have to move away from SFF systems.

[leadeater]

12 minutes ago, yian88 said:

It doesnt exist,  6 core HEDT i7 7800X socket 2066 is slightly more expensive than socket 1151v2 i7 8700k, because intel wont cannibalize their mainstream platform by selling i7 7800x cheaper than 8700k so you can figure out the iGPU is actually quite an expensive piece.

Add the fact that the cheapest 2066 board i can find is 2-3x the price of a cheap b350 1151v2 board.

 

Basically if i want a performant i5 8 core unlocked without iGPU and fancy HEDT features like quad channel RAM, and dual GBE ports (which is overkill for regular user) intel will never give that to the market, they are sandbagging performance as i said. They will keep this model for as long as possible, unless AMD can deliver some serious competition with 8-16 cores on AM4 socket then i hope intel might have no choice but to deliver high end sku's i5/i7/i9 with 8-10-12 cores by removing iGPU, we will see when new i9 for 1151v2(or v3?) comes out, thats when i will know not even over their dead company they will change this scam business model ( i think so).

Thing is there is no way DDR4 dual channel can service 10 or 12 cores, it already struggles with 4 cores and AVX2. Even ignoring AVX 10 or 12 cores is just too much for dual channel if you want to effectively use them. There is a reason why the older X58/LGA1366 Xeons were triple channel. When DDR5 comes out then you could do that many cores at dual channel.

 

If it were as simple as chucking out the iGPU and throwing more cores in it would have happened. I know a lot of what Intel does looks bad but there are some good technical reasons behind some things, I don't like many things Intel does but there are reasons and not just to screw money out of people.

 

And you can't just say it doesn't exist, it does and it's the HEDT line and as I suspected you just don't like the price of that platform. You have reasons to have that opinion but it exists either way. An Intel product line with no iGPU and more cores.

[yian88]

21 minutes ago, leadeater said:

Thing is there is no way DDR4 dual channel can service 10 or 12 cores, it already struggles with 4 cores and AVX2. Even ignoring AVX 10 or 12 cores is just too much for dual channel if you want to effectively use them. There is a reason why the older X58/LGA1366 Xeons were triple channel. When DDR5 comes out then you could do that many cores at dual channel.

 

If it were as simple as chucking out the iGPU and throwing more cores in it would have happened. I know a lot of what Intel does looks bad but there are some good technical reasons behind some things, I don't like many things Intel does but there are reasons and not just to screw money out of people.

 

And you can't just say it doesn't exist, it does and it's the HEDT line and as I suspected you just don't like the price of that platform. You have reasons to have that opinion but it exists either way. An Intel product line with no iGPU and more cores.

Im not buying that for a second unless you have some well documented/tests source for that. Im not saying dual channel was ever enough for certain tasks.

Just like first 8 core intel was an i9 HEDT and now they move it to mainstream dual channel RAM 8 core i9 9900k, i have no doubts 8-12 cores can run without issues with dual 3000+ mhz RAM for most  workstation tasks, sorry i have not documented on this issue so i dont know specifically for intel 14 nm kaby lake cores, how much bandwidth they use in various tasks, i would assume if you throw work 100% on 12 cores at the same time with intensive memory read/write dual channel is not enough but otherwise for workstation/multitasking it should be enough. By that i mean general software like CAD or photoshop or streamers/gamers dont need more than 2 channel for now.

 

Regardless that is not a reason to hold back on making new chips without iGPU and more cores, 8 core i5, 10 core i7/i9 is totally feasable on dual channel, probably new socket would be required.

This is an uninformed opinion but if you can run 6 cores i7 at 4.5ghz+ and the iGPU off the same dual channel ram (GPU's are bandwidth hungry) your theory of dual channel not being enough for more than 6-8 cores doesnt stand, that or the 6 core i7+ iGPU doesnt have enough bandwidth already on 1151v2 and its choking itself when both are under load, which means intel has been selling cpu+gpu combo that has too little bandwidth and cant be used to their full potential.

 

If the upcoming core i9 8 core with fat L3 and 5ghz has iGPU all running of dual channel ram, then its true that they have enough bandwidth for 8-12 cores or more with dual channel, maybe not all software will perform at their peak on dual vs quad but it will be enough for most tasks.

 

Even if a product line like HEDT with no iGPU exists they are not priced accordingly due to the reasons i mentioned, and i do not care about the concept of intel cpu withouth iGPU, i think from my own analysis over the years that intel is sandbagging performance because of that iGPU on mainstream since they still have performance superiority, and i do not buy above mainstream platform because i do not have HEDT needs asside from more cores, in your view i should buy  the 1.5-2x more expensive HEDT cpu+motherb combo  if i want 2+ more cores, in  my view they should release new i5/i7 line with no iGPU but more cores for the same price as i5/i7 with iGPU, its a big difference, its all about bang for buck and business model.

I also desire all intel chips with unlocked multiplier on all boards. There is just too much scum business decisions at intel, thats why they make so much profit.

 

Im not saying AMD is a saint, they are in a shitty situation and making business decision as intels would imediately bring them close to bankrupcy again. At this point AMD can only make honest cpu products, unlocked, many cores, decent performance no bullshit, and thats what i want from intel aswell, to take off their coat of superiority and give us some decent honest products.

[thorhammerz]

2 hours ago, leadeater said:

And you can't just say it doesn't exist, it does and it's the HEDT line and as I suspected you just don't like the price of that platform. You have reasons to have that opinion but it exists either way. An Intel product line with no iGPU and more cores.

B-b-b-but, why can't I have everything I want, at the price-point I want (or hell, just make it free)?! 

 

/s 

[79wjd]

28 minutes ago, thorhammerz said:

B-b-b-but, why can't I have everything I want, at the price-point I want (or hell, just make it free)?! 

 

/s 

Fuck that, my time is expensive; they should be paying me to honor them with my use of their product.

[mynameisjuan]

13 minutes ago, djdwosk97 said:

Fuck that, my time is expensive; they should be paying me to honor them with my use of their product.

I dont think you get how this works.

[Brooksie359]

5 hours ago, AngryBeaver said:

Because power consumption isn't a big deal for me. I am not running 100's of servers where I need to care about energy costs at any meaningful level. You don't see this happening on the Xeon products.

 

10-15% clock speeds PER core is a pretty big thing. If we look at a 6 core cpu that is 6x14 = %84 more compute power before even looking at things like IPC.

I am not comparing stock to OC. It just looks that way, because if you buy say a 2700x and try to OC it.. you MIGHT get 100-150mhz more. On an intel chip at 4.7ghz you can easily push them to 5ghz or more with proper cooling.

 

Single Thread performance will never be ancient garbage. Even with current gen titles you can't just hit one magic button and have it automatically split the load between the available cores. You have to code in how it splits the work load which changes depending on the amount of cores you have to access. Some items cannot even be split and must depend on running on a single core or they are high usage actions that would occupy a single core regardless.

 

Turbo and XFX are no different than turbo-boost though... so I am not sure why you are tossing that out there.

Yes, DX12 does help with the problem, but it still is a problem which is why you still see higher fps with intel cpu's on even DX12 games. The problem will be less apparent too as people shift to higher res gaming, but then again once 4k becomes the norm and the GPU tech catches up to that we will see similar problems to what we have now at 1080p gaming. I mean who cares if you are getting 160fps instead of 190, but it does give you an idea of how much better intel cores are.

I am sure AMD will improve their IPC, I count on it. Honestly, they should have an easier go on it since the architecture is so new.

 

Yes they did gain some performance going to Ryzen+, but that isn't a great feat. Intel has still managed to get 3-5% ipc improvements on their tic cycles. Then don't forget intel has a new node coming out in about 2 years. If you don't think 10nm and 10nm+ and 10nm++ won't be huge jumps then you don't know intel. Intel has a huge R&D budget and plenty of budget to push things early if they had to. So the fact they are willing to wait till 2020 tells me they are not that concerned with the new 7nm stuff (which once again there is no standard for measuring it might be closer to a 10nm process compared to how intel measures their process)

 

Except that is only a guess and a very very rough one with no data to back it. I will be very surprised if they hit even 4.7 ghz on 7nm.

 

Do you think they are the only ones that can toss more cores on a chip? Intel has the same capability they just choose not to and part of me things they are using it as a way to milk us for more money. They can just up the core count slowly over each new generation and we will keep wanting to have the latest and greatest so we just throw them more money. Maybe not the best thing for us consumers, but smart business for intel.

 

It doesn't look good nor bad. It is too early to tell and Intel has shown they can pull some tricks out of their sleeve when needed. Don't get me wrong, I am all for AMD gaining an advantage it makes it a much more competitive market for us consumers and it is the only real time we see leaps in technology. Look how stale the CPU world was while AMD was running the old bulldozer crap.

Efficiency is hard to give an accurate representation of.. they trade blows. You have to look at everything including instruction sets and the like. I will say though AMD is doing extremely well in highly multi-threaded efficiency.

 

Heck we have had some really good over-clockers in the past decade. Despite being terrible the old bulldozer stuff could overclock well and on the intel side you had sandybridge which also overclocked extremely well. Taking a base block 2600k of 3.4ghz to 5ghz(about 47%) wasn't unheard of or even that hard provided you had a decent custom loop.

 

 

 

 

15% higher clocks means 15% higher overall performance. The i7 8700k has around 25% higher clocks while the 2700x has 33% more cores. By your metric 6×25= 150% higher performance meaning the 8700k should preform way better in multicore performance but it doesn't. 15% is 15% regardless of the amount of cpu cores you have. Don't try and make Intel's performance gap due to clock speed seem like more than they are. They are around 20 to 25% which isn't insignificant but also not the 84% you are talking about. 20 to 25% clock advantage isn't impossible to close with smaller nodes like 7nm so that gap will likely shrink and we will see what happens then. Anyways I just wanted to explain why your reasoning is wrong. 

[AngryBeaver]

31 minutes ago, Brooksie359 said:

15% higher clocks means 15% higher overall performance. The i7 8700k has around 25% higher clocks while the 2700x has 33% more cores. By your metric 6×25= 150% higher performance meaning the 8700k should preform way better in multicore performance but it doesn't. 15% is 15% regardless of the amount of cpu cores you have. Don't try and make Intel's performance gap due to clock speed seem like more than they are. They are around 20 to 25% which isn't insignificant but also not the 84% you are talking about. 20 to 25% clock advantage isn't impossible to close with smaller nodes like 7nm so that gap will likely shrink and we will see what happens then. Anyways I just wanted to explain why your reasoning is wrong. 

You are off on your math.. for a 6 core to out perform the 8 core the 150% would still be short. It would take 200% overall performance to = the 8 core in multi-threading. The reason is 100% more performance is per core... so 100% = 1 additional core basically. So a 6 core with 200% overall core performance would be = to the 8 core AMD.

 

I hope my explanation makes sense.

 

Lets look at this from a pure computing power perspective

 

4.3 x 8 = 34.4 

4.7 x 6 = 28.2

 

28.2/34.4 = 29% different in compute power.

 

From http://hwbench.com/cpus/amd-ryzen-7-2700x-vs-intel-core-i7-8700k

 

Multi thread performance index

• AMD Ryzen 7 2700X
  129%
• Intel Core i7-8700K
  100%

Relative performance

 

So pretty much spot on.

 

 

The problem comes from the fact the 2700x doesn't really OC past 4.3ghz  The 8700k though can hit 5.2ghz with a potential delid and decent cooling solution.

 

So that changes the situation because of the OC ability of intels

 

Now the equation looks like this

 

4.3 x 8 =34.4

5.2 x 6 =31.2

31.2/34.4 = 10% difference in overall compute power.

 

Now my point is the 8700k has 2 less cores... so 33% less cores, but only 10% less overall power...

 

So from your comparison we can use the core to core power example   4.3/5.2 or 18% more power per core X 6 cores is 108%

 

So that means we basically gained the power of a little over 1 AMD core by just the frequency... the rest can be account for by the IPC advantage which I didn't even touch here.

[The Benjamins]

4 minutes ago, AngryBeaver said:

You are off on your math.. for a 6 core to out perform the 8 core the 150% would still be short. It would take 200% overall performance to = the 8 core in multi-threading. The reason is 100% more performance is per core... so 100% = 1 additional core basically. So a 6 core with 200% overall core performance would be = to the 8 core AMD.

 

I hope my explanation makes sense.

 

 

27 minutes ago, Brooksie359 said:

15% higher clocks means 15% higher overall performance. The i7 8700k has around 25% higher clocks while the 2700x has 33% more cores. By your metric 6×25= 150% higher performance meaning the 8700k should preform way better in multicore performance but it doesn't. 15% is 15% regardless of the amount of cpu cores you have. Don't try and make Intel's performance gap due to clock speed seem like more than they are. They are around 20 to 25% which isn't insignificant but also not the 84% you are talking about. 20 to 25% clock advantage isn't impossible to close with smaller nodes like 7nm so that gap will likely shrink and we will see what happens then. Anyways I just wanted to explain why your reasoning is wrong. 

ok at stock: 2700x is 24% faster then a 8700k ( clock x cores)

2700x 8c @ 4.0 = 32

8700k 6c @ 4.3 = 25.8

 

OC: 2700x is 7.7% faster then a 8700k ( clock x cores)

2700x 8c @ 4.2 = 33.6

8700k 6c @ 5.2 = 31.2

 

Rumor: 9900k is 17.5% faster then a 2700x ( clock x cores)

9900k 8c @ 4.7 = 37.6

9700k 8c @ 4.6 = 36.8

[Brooksie359]

17 minutes ago, AngryBeaver said:

You are off on your math.. for a 6 core to out perform the 8 core the 150% would still be short. It would take 200% overall performance to = the 8 core in multi-threading. The reason is 100% more performance is per core... so 100% = 1 additional core basically. So a 6 core with 200% overall core performance would be = to the 8 core AMD.

 

I hope my explanation makes sense.

 

200% performance is 2 times the performance. You said 15% clock increase mean 84% more compute power which it doesn't. 15% increase in clocks means 15% increase in compute power. You can't just use percentages how you like. If someone says this cpu has 150% higher compute power that means you would simply multiply the other cpus compute power  by 1.5 and it should give you the compute power of the higher performing chip. So based on what you said you are simply wrong. If you had worded it differently maybe what you said would make more sense but in its current form it is incorrect. 

[Brooksie359]

4 minutes ago, The Benjamins said:

 

ok at stock: 2700x is 24% faster then a 8700k ( clock x cores)

2700x 8c @ 4.0 = 32

8700k 6c @ 4.3 = 25.8

 

OC: 2700x is 7.7% faster then a 8700k ( clock x cores)

2700x 8c @ 4.2 = 33.6

8700k 6c @ 5.2 = 31.2

 

Rumor: 9900k is 17.5% faster then a 2700x ( clock x cores)

9900k 8c @ 4.7 = 37.6

9700k 8c @ 4.6 = 36.8

Yeah I was making a rough estimate. I wasn't really trying to be precise just trying to explain why his comment on the significance of clock speed difference is wrong. 

[The Benjamins]

1 minute ago, Brooksie359 said:

Yeah I was making a rough estimate. I wasn't really trying to be precise just trying to explain why his comment on the significance of clock speed difference is wrong. 

ya, their math made my head hurt.

 

I will add my Ryzen 3000s speculation

 

3700x 10c @ 4.0 = 40 (6.3% faster then 9900k) (same boost as 2700x)

3800x 12c @ 3.7 = 44.4 (18.1% faster then 9900k) (same boost as 1920x)

[79wjd]

18 minutes ago, AngryBeaver said:

You are off on your math.. for a 6 core to out perform the 8 core the 150% would still be short. It would take 200% overall performance to = the 8 core in multi-threading. The reason is 100% more performance is per core... so 100% = 1 additional core basically. So a 6 core with 200% overall core performance would be = to the 8 core AMD.

 

I hope my explanation makes sense.

 

Your numbers are off. An 8700k needs a 33% single core performance lead to equal a 2700x in multithreaded, perfectly scaling, loads. It might be easier to see if you use Ryzen as the baseline.

 

E.g.

 

let 1 Ryzen core at 4.2ghz be 1 unit of performance. Then a 2700x would be 8 units powerful.

 

An 8700k at 5.2 has a 24% clock speed advantage and let's just say a 5% IPC lead at the same clocks (I don't know what the IPC difference is with Ryzen 2). So then 1 8700k core would have 1*1.24*1.05=1.30 units of power. Multipled by 6 gives you 7.8 units -- just shy of a 2700x.